LG Uplus Targets B2B Market With Token Optimization to Cut AI Agent Costs

LG Uplus plans to enter the B2B market with token optimization as its entry point, aiming to sharply cut the cost of running AI agents, according to a Chosunbiz report.

Tokens are the basic unit of text that models process and, for most commercial models, the basis of billing. Agents put far more pressure on that meter than simple chat applications: completing a task often requires multiple rounds of reasoning, calls out to external tools, and sending tool results back into the model, so token consumption multiplies with every step.

That makes token optimization a commercial variable rather than a technical footnote. If the cost of a single task cannot come down, enterprises will struggle to move agents out of pilots and into production in high-frequency settings such as customer service and internal process automation.

There is a logic to a telecom operator entering this layer. Carriers already own networks, data centers and relationships with enterprise customers, which lets them move from selling connectivity and compute toward selling model access and agent runtime on top. LG Uplus is using token efficiency as its selling point, essentially trying to differentiate on cost and engineering efficiency in a market where model capability is converging.

The report, however, offers only directional language. It does not disclose the product form, pricing model, target industries or existing customers, and those details will determine whether this is an add-on bundled with cloud and connectivity services or a standalone agent platform.

Three things are worth watching: the concrete method behind the token optimization and any verifiable cost reduction, whether the offering targets Korean enterprises or a broader regional market, and whether carrier-led agent services can hold their ground against model vendors and cloud providers selling into the same accounts.